[Nexus-developers] NeXus project plan, second draft

[Ray Osborn]

Mark,
Sorry not to have participated in these discussions for a while, but I  
think this is an excellent document that should help to clarify in our  
minds just what the NeXus priorities should be.  I certainly  
appreciated the candid critiques of what has been happening at the  
NIAC, and why progress is sometimes so slow.

Here are a few comments:

1) Organization

The fact that everyone is still working on NeXus as a 'volunteer',  
even when the format is the declared standard for the facility and  
most of the NIAC members have software development as part of their  
job description, is troubling.  It suggests that we have not done  
enough to persuade our management that NeXus is now a critical part of  
their mission, and we have perhaps not persuaded ourselves that we  
should be committing more of our own time (myself included, although I  
do virtually no software development now).  We have spent so long  
doing this in our spare time that we have got used to it, and haven't  
realized that we should now be arguing for it to be otherwise.  There  
are exceptions, of course.  Much of NeXus early progress was due to  
the fact that people like Mark committed themselves to getting the API  
completed, but we need to find a way for NeXus to become part of more  
people's day jobs.

I suggest that we revisit the idea of seeking funding for NeXus from  
the facilities, at least in the area of travel support to meetings.   
Attendance in Australia was low for this very reason.  A facility  
subscription to NeXus will ensure that management will pay attention  
to ensure that they are getting value for their money.  I heard a  
worrying report that the management at one major facility, closely  
involved in NeXus development, has concluded that it is no longer  
worth supporting NeXus.  I think that happened because of this  
disconnect, and we don't want it to become part of a trend.

A couple of NIAC meetings ago, I talked about efforts that we are  
taking at Argonne to try and secure more funding for computational  
issues at facilities through an institute known as ASISI.  Those  
efforts had been stalled because of stagnant budgets at the DOE, but  
there is a realistic prospect of a better funding climate with the new  
administration, so I will keep you posted if we get any positive  
news.  If ASISI does get funded, then I would propose, at least  
temporarily, to become the ASISI representative on the NIAC, so  
solving the problem of what to do with dead facilities like IPNS.   
Perhaps the constitution would have to be amended to allow members to  
represent facilities and/or scattering institutes.

2) NeXus Software

Since I have had some problems installing NeXus on my machine, I might  
be behind the times here, but I think another key application would be  
a simple GUI browser, i.e., a GUI version of NXbrowse with simple  
visualization added.  I think if this were a standard part of the  
distribution, it would help us explain the value of NeXus to others.   
There was some work in this area a few years ago, but it seems to have  
stalled.  Also, if we ever got more software resources, a Web 2.0  
interface would be useful for institutes with catalogued data online.

I also would reiterate my plea that we simplify NeXus installations,  
since the biggest barrier-to-entry would be an inability to install  
NeXus in the first place.  It needs to be trivial.  I'll see if I can  
find someone at Argonne to help with Mac OS X, but we need to have a  
systematic way of addressing holes in our operating system support.

3) Processed Data

I haven't thought about this as carefully as I should, but can't we  
just define processed data as

<NXentry>
   <NXsample>...</NXsample>
   <NXdata>
     e.g. S(Q) or S(Q,w) or whatever vs Energy or Q or whatever
    </NXdata>
</NXentry>

and then declare victory and go home?  The long title of the data  
items  and the axes would define what the data contained.  In  
principle, since this is processed data, we don't need any  
instrumental data, except perhaps a link back to the original  
NXentry.  We could also have multiple NXdata items for non-contiguous  
patches of (Q,w)-space.  I presume that is what you meant in your last  
sentence.

4) Opportunities

I agree wholeheartedly with the idea of seeking active collaboration  
with DAVE, DANSE, etc, to make sure that NeXus is embedded in their  
products in a sensible way.

With best regards,
Ray

On Mar 10, 2009, at 10:50 AM, Mark Koennecke wrote:

> Hi,
>
> find attached the second draft of the NeXus project plan. I did not  
> yet get any feedback from Stephane
> on NXtv.
>
> Best Regards,
>
>  Mark
>
> <NXproject.pdf>_______________________________________________
> NeXus-developers mailing list
> NeXus-developers at nexusformat.org
> http://lists.nexusformat.org/mailman/listinfo/nexus-developers

-- 
Ray Osborn
Materials Science Division
Argonne National Laboratory
Argonne, IL 60439, USA
Phone: +1 (630) 252-9011
Email: ROsborn at anl.gov